The stability of Cl-, Br-, and I-passivated Si(100)-(2 × 1) in ambient environments for atomically-precise pattern preservation

Feature size miniaturization is a driving force in the development of increasingly small, low-cost, high-powered computing devices. As features approach sub-10 nm, new challenges for miniaturization arise. State-of-the-art atomic precision advanced manufacturing (APAM) can be used to make devices with atomically-sharp features and unprecedentedly high doping concentrations [1]. Atomic precision is achieved by using a scanning tunneling microscope (STM) tip to selectively remove H from a H-passivated [2] or halogens from a halogen-passivated [3] Si substrate, to expose Si dangling bonds. The high reactivity of the Si dangling bonds relative to passivated Si enables selective reactions within depassivated areas with molecular precursors [4]. This is conceptually similar to area-selective atomic layer deposition [5, 6], but in APAM, the use of an STM tip for lithography enables atomic-level patterning precision in the lateral dimensions.

APAM processes to date have been inherently confined to ultra-high vacuum (UHV) environments due to their dependence on Si dangling bonds which do not survive outside of UHV. This UHV requirement is a strong deterrent to the implementation of APAM technologies in commercial device fabrication. To extend APAM processing beyond UHV environments, the Si dangling bonds must be passivated to protect against oxidation while still maintaining a difference in chemical reactivity as compared to the surrounding resist area. UHV-prepared H-passivated Si(100) is known to survive extended periods of time in ambient environments [7–9] whereas wet-prepared H-passivated Si(100) will oxidize after several hours [10]. However, the stability of halogen-passivated Si(100) has not been well explored beyond Si passivated with Cl [11–13]. Halogens show a promising potential as a passivation layer for pattern preservation as Ferng et al. [14] have previously demonstrated the formation of local Cl- and I-passivated patterns on an otherwise H-passivated surface. While these previous results set an important precedent in developing a co-patterning capability, the unknown stability of these chemistries and patterns outside of UHV environments limits their use as a means to integrate APAM Si samples into real-world applications.

As previous work has suggested that Si(100) surfaces passivated with Cl are not stable in ambient [11, 15–17], we investigated the stability of UHV-prepared halogen-passivated Si(100) (halogen-Si, specifically Cl–Si, Br–Si, and I–Si) as well as H-passivated Si(100) (H–Si) in technologically relevant environments including N₂ and ambient air. Using STM combined with x-ray photoelectron spectroscopy (XPS) we found that Si(100) surfaces fully passivated with H, Cl, Br, and I were all fully stable within an inert N₂ environment for at least 44 h. Moreover, all but Cl–Si were highly resistant to degradation/oxidation in ambient conditions for at least 8 h. While the ambient stability of H–Si was already known, the additional discovery that Br–Si and I–Si remain intact in ambient conditions enables the preservation of APAM patterns outside of UHV. As a proof-of-principle demonstration of pattern preservation, several regions were lithographically patterned on a H–Si sample with patterns sequentially passivated with Cl, Br, and I before being exposed to N₂ and ambient. All the halogen regions maintained their integrity post-exposure to N₂, and the Br–Si and I–Si regions continued to maintain their integrity post-exposure to ambient. These results provide the foundation for the use of atomically-precise patterning on technologically relevant Si(100) in processes beyond the confines of UHV systems for integration with additional selective chemistries and commercial device processing tools.

Sample preparation and STM imaging were performed in a UHV system equipped with a Scienta Omicron VT-STM and ZyVector STM lithography control system. The base pressure of the sample preparation chamber ranged from P < 2.7 × 10⁻⁹ Pa (2.0 × 10⁻¹¹ Torr) to P < 1.1 × 10⁻⁸ Pa (8.0 × 10⁻¹¹ Torr) depending on the experiment, while the pressure in the STM chamber was P < 1.3 × 10⁻⁹ Pa (1.0 × 10⁻¹¹ Torr). Si(100) wafers were obtained from Prolog and ITME. They were p-type, B-doped with resistivities ranging from ρ = 0.001 Ω cm to ρ = 10 Ω cm and were oriented within 0.5° of (100). The Si samples were cut to 4 mm × 12 mm in size, cleaned by sonication in acetone, methanol, and isopropanol, mounted on a Scienta Omicron XA sample plate, and loaded into the UHV chamber. Clean Si(100)-(2 × 1) surfaces were prepared by flash annealing to 1200 °C following the procedure found in reference [18].

To produce halogen–Si surfaces, freshly prepared clean Si samples were exposed to a molecular halogen flux at 200 °C to 250 °C in UHV to minimize water contamination [19] and prevent accumulation of inserted halogen species [20–24]. Cl₂, Br₂, and I₂ were generated from solid-state, electrochemical cells consisting of AgX (X = Cl, Br, or I) doped with 5 wt.% of the corresponding cadmium halide for Cl₂ and Br₂ [25], while RbI was used for I₂ [26]. After exposure, the halogen–Si samples were further annealed (200 °C for I, 370 °C for Br, 425 °C for Cl) to remove physisorbed halogens [27]. These annealing temperatures were chosen to be low enough to prevent activation of surface etching and roughening processes [28, 29]. During halogen exposure, the pressure rose by no more than 1.3 × 10⁻⁹ Pa (1 × 10⁻¹¹ Torr) above the base pressure. H–Si samples were prepared by exposing cleaned Si surfaces to a flux of atomic H generated by a H atom beam source (MBE Komponenten) with the substrate held at 350 °C.

For STM-based studies of the stability of bare Si and halogen–Si surfaces, samples were scanned at the beginning of the experiment to determine the baseline defect concentration. Defect coverages are reported in monolayers (ML), where 1 ML is equivalent to the dangling bond density of the Si(100)-(2 × 1) surface (6.78 × 10¹⁴ cm⁻²). Gwyddion [30] was used for image processing to determine the degree of change of the surface after samples were exposed to various conditions. To quantify defect coverages before and after exposures, a threshold was set to include all features on the surface that appeared significantly darker than a nominal surface dimer height. This thresholded surface area, reported as the defect coverage in ML, was compared for multiple pre- and post-exposure areas of the surface and averaged to determine the change in the defect coverage. By doing this, the baseline defect concentration of the initial surface including dimer vacancies was subtracted off so the result reflected the accumulation of defects on the surface with exposure time. Uncertainties in the defect coverage were determined by calculating the standard error of each data set arising from statistical fluctuations in the local defect coverage on different areas of the surface. The change in defect coverage at various time points was used to determine a defect accumulation rate by fitting a line to the data (see supporting information, or SI (https://stacks.iop.org/JPCM/33/444001/mmedia)) with uncertainty in the rate determined by the fit uncertainty.

To study the effect of a UHV environment on the various surfaces, samples were left within the STM chamber with the tip retracted for various exposure times before the surface of each sample was scanned again. To expose the bare Si, H–Si, and halogen–Si surfaces to an inert N₂ environment, samples were placed in the UHV system load lock (base pressure P < 1.3 × 10⁻⁶ Pa (1.0 × 10⁻⁸ Torr)) after surface preparation and initial STM image acquisition. The load lock was then backfilled with dry, filtered N₂ to approximately 1 atm (101.3 kPa). Care was taken not to have nitrogen directly impinging upon the sample surface during backfilling. Samples were left in this environment for a given period of time as described in the text after which the load lock was evacuated. Samples were transferred back into the UHV system after about 1 h of pumping down to a pressure of roughly P = 6.7 × 10⁻⁵ Pa (5.0 × 10⁻⁷ Torr).

After introduction back in the UHV system, samples were annealed to between 150 °C to 350 °C to remove physisorbed species before transfer to the STM for scanning. The annealing temperature was modified for each resist to be below the desorption temperature of the respective H or halogen. Annealing was necessary because it became difficult to maintain controllable imaging conditions due to frequent changes in the STM tip caused by interactions with loosely-bound species on the surface. We attributed these loosely bound species to physisorbed water molecules [31]. These species were readily removed with a mild anneal, after which it was once again possible to obtain clear images. Samples exposed to 2 h of N₂ were also analyzed using in situ XPS measurements within the UHV system before annealing. The in situ XPS spectra were collected using a Mg anode x-ray source with a DESA 100 cylindrical mirror analyzer, both from Staib Instruments.

To study the exposure of the bare Si, H–Si, and halogen–Si surfaces to ambient air, samples were removed from the UHV system via the load lock into the environment of the lab and placed under a glass dish to prevent dust or particle accumulation. For subsequent STM analysis, samples sat in ambient for a given length of time before being loaded back into the load lock for pump down and re-introduction to the UHV chamber. Again, samples were lightly annealed before STM scanning. Samples exposed to ambient were analyzed using ex situ XPS measurements. Samples with long exposure times were transported in ambient to the XPS, while samples with shorter exposure times were exposed to ambient in the lab prior to being loaded into an N₂ environment for transport. Samples that were kept under N₂, were loaded in the XPS intro chamber within 30 min of N₂ exposure. XPS was collected on a Kratos Axis 165, operating in hybrid mode, using monochromatized Al Kα x-rays (240 W). Survey spectra and high resolution spectra were collected at a pass energies of 160 eV and 40 eV respectively. Data was collected at a take-off angle of 20 degrees with respect to the sample surface, and the iris was reduced to improve angular resolution.

The stability of halogen–Si and bare Si surfaces in various environments is studied. In the context of this discussion, we define a stable surface as one that does not readily react with its environment, maintaining its initial surface structure and specifically does not form more than 0.01 ML of surface oxide, although this goal would be application dependent. These surface stabilities have not been well explored previously despite being relevant for compatibility with the semiconductor industry. This is important to investigate as the ability to convert the Si dangling bonds into a robust and ambient stable chemistry would provide a much-needed route for bringing APAM processes out of the confines of UHV. These studies on Si(100) are especially important as Si(100) reactivity is less explored than the well-understood Si(111) and Si(100) reactivity differs from Si(111) reactivity. For example, controlled exposure of Cl-passivated Si(111) and Cl-passivated Si(100) to D₂O has demonstrated that the Cl-passivated Si(111) reacts more readily than Cl-passivated Si(100) with D₂O species [12].

Figure 1 depicts representative filled-state STM images of (a1) bare Si(100), (b1) Cl–Si(100), (c1) Br–Si(100), and (d1) I–Si(100) as prepared under UHV conditions. Due to similarities in their appearance within STM images, in situ XPS spectra of the Cl 2p, Br 3d, and I 3d are shown as insets in figures 1(b1), (c1) and (d1), respectively, as verification of the chemical identity of each halogen passivation shown. Survey scans for all initial surfaces are provided in the SI. As shown in figure 1, the Si surfaces appear fully passivated with single halogen atoms passivating each Si dangling bond, maintaining the underlying (2 × 1) pattern. Dark features spanning a dimer row are Si dimer vacancies present before passivation while smaller dark features observed on half a dimer row are likely single missing halogen atoms [32, 33]. Brighter features can result from physisorbed halogens. Typically, only a small number of unpassivated dangling bonds remain present when preparing surfaces in this manner, demonstrating that near full passivation is easily achieved for all halogen passivation layers. Moreover, it is clearly evident that Br and I form fully packed MLs on Si(100), despite their larger size and consequent increased steric hindrance. Starting surfaces of samples used in stability studies described here werequalitatively similar to those shown in figures 1(a1), (b1), (c1) and (d1).

Zoom In Zoom Out Reset image size

3.1. Halogen passivation stability in UHV

3.2. Halogen passivation stability in N₂

The stability of the halogen passivation layers was studied in an inert N₂ environment using a combination of in situ XPS and STM characterization. Fully-passivated samples were moved from the UHV system to the load lock, which was subsequently filled with dry, filtered N₂ to a pressure of approximately 1 atm. After times ranging from 1 h to 44 h, the load lock was pumped out and the samples were reintroduced to the UHV environment. Each sample was examined in situ with XPS before and after exposure and compared with the Si 2p spectra of an unpassivated Si sample that was freshly prepared in UHV and exposed to similar conditions within the load lock. We use the Si 2p region to analyze the stability of the samples with XPS. Oxide growth occurring from resist degradation appears as a distinct peak between 102 eV and 104 eV and is a direct measure of chemical bonds between Si and O. Figure 1(f) displays the XPS spectra of the Si 2p region for each sample, with the inset showing a magnified view of the oxide region. For the bare Si sample, we observe a slight increase in the spectrum at 102 eV to 104 eV, which suggests the formation of some SiO_x , as expected for an unpassivated sample. No increase was observed for the passivated surfaces. While high purity, filtered N₂ was used to backfill the load lock, trace amounts of water are always present and will bond to highly reactive bare Si [35].

Due to the low peak intensity shown in the figure 1(f) inset, we analyzed STM images of the various surfaces to corroborate these stability results. STM images showed the halogen–Si surfaces held up remarkably well relative to bare Si, with no noticeable changes in the surface structure after N₂ exposure. Figures 1(b2), (c2) and (d2) show filled-state STM images of the Cl–Si, Br–Si, and I–Si surfaces, respectively, shown in (b1), (c1), and (d1) after exposure to N₂ for 44 h. The halogen-passivated surfaces appear largely similar to the initial surfaces with halogen passivation layers clearly intact. While a double tip artifact and some smearing is evident on the I–Si image in (d2), dimer rows are still visible indicating the surface structure with I is still intact. However, a slight increase in the number of dark, dimer-sized defects is observed in some images. These halogen-passivated surfaces were found to be remarkably stable in N₂, with very small defect accumulation rates of 0.007(2) ML/day for Cl-Si, 0.001(2) ML/day for Br-Si, and 0.004(4) ML/day for I–Si, as shown in figure 1(e). As was the case with the defect accumulation rates for the halogen-passivated surfaces in UHV, these N₂ rates for Br–Si and I–Si are small relative to their uncertainty and thus are not statistically above zero based on these measurements. STM images of Cl–Si, Br–Si, and I–Si surfaces exposed to N₂ for 2 h are also shown in the SI. As a comparison, a H–Si sample was also prepared and exposed to N₂ for 2 h. STM showed the exposed surface appeared qualitatively similar to the initial surface, much like the halogens (see SI). The change in the defect coverage after exposure was determined to be 0.014(5) ML after 2 h, which is larger than what was seen for Cl–Si and Br–Si after 2 h but similar to one I–Si sample. We note this single time point is not statistically sufficient to determine a long-term accumulation rate, but as H–Si is known to be stable in air for some time [7], it reasons that it is similarly stable in N₂. Meanwhile, in stark contrast to the halogenated surfaces, the bare Si surface showed obvious degradation after exposure to N₂ for 2 h, as seen in figure 1(a2). Numerous dark spots are visible on the dimer rows indicating chemisorbed species covering a larger portion of the surface. The calculated defect accumulation rate for the bare Si sample in N₂ was 3.1(4) ML/day, or roughly three orders of magnitude larger than for the halogens, as shown in figure 1(e) (note the vertical axis break). These results for N₂ stability agree with previous observations that Cl–Si(111) [11] and Br–Si(100) [36] surfaces are resilient in N₂ environments.

While in situ XPS data showed that all passivated samples had equal resistance against chemisorbed water in N₂, we observed more physisorbed water on the H–Si surfaces than the halogen–Si surfaces. After 2 h of N₂ exposure, each sample showed an increase in O species, while C was not observed in the survey scans or C 1s regions. Figure 2 shows XPS spectra of the O 1s region normalized by Si 2p peak height of the samples after 2 h in N₂. O levels were observed to be the highest on the bare sample, as expected. The O levels on the bare Si are attributed to both physisorbed and chemisorbed O species, since the bare Si sample exhibited c-type defects from chemisorption of dissociated water under N₂. Since the H- and halogen-passivated samples were found to be stable in N₂, the O on passivated samples was attributed solely to physisorbed species. The ratio of the O 1s area to Si 2p area for each surface is listed in the table of figure 2. The O signal on the halogen–Si samples was roughly half that on the H–Si surface, with relative peak heights Cl < I < Br < H < bare Si, which suggests that halogen-passivated surfaces have a lower probability for physisorbing water than H–Si. For lithographically defined patterns on H–Si or halogen–Si, physisorbed molecules may diffuse along the surface and can interfere with patterning by integrating into bare Si sites. A lower concentration of physisorbed water on halogen surfaces suggests that patterned areas on halogen–Si may have longer UHV stability than similar patterns on H–Si.

Zoom In Zoom Out Reset image size

3.3. Halogen passivation stability in ambient

To test the ambient stability of the halogen passivation layers, each sample was exposed to the laboratory ambient environment for varying times between 15 min and 8 h. Ex situ XPS spectra of the Si 2p region for these samples are shown in figure 3(a) (1 h) and figure 3(b) (8 h), with insets showing magnified SiO_x regions. The H–Si, Br–Si, and I–Si samples were found to be stable after exposure to ambient for 1 h with no detectable SiO_x formation, as shown in the figure 3(a) inset. Furthermore, Br–Si and I–Si surfaces were remarkably resistant to degradation up to 8 h in ambient, as indicated by the spectra shown in the figure 3(b) inset. After 8 h in ambient, the Br–Si spectra shows only a small amount of oxidation with the ratio of the SiO_x area to the total Si 2p area, R_ox, being just 0.020, and the I–Si spectra exhibits effectively no detectable SiO_x peak (R_ox = 0.006). While 8 h was the longest time studied here, we presume these surfaces to remain relatively intact for far longer periods of time, similar to H–Si, which was previously shown to be stable for at least 42 h in ambient [9].

Zoom In Zoom Out Reset image size

After 1 h in ambient, the Cl–Si sample was the only passivated sample to exhibit a significant SiO_x peak (R_ox = 0.105) as seen in figure 3(a) inset, and additionally, a small SiO_x peak was observed on Cl–Si (R_ox = 0.015) after just 15 min (see SI). Surprisingly, after 1 h exposures in ambient, the intensity of the SiO_x peak was larger on Cl–Si (R_ox = 0.105) than on bare Si (R_ox = 0.090). This was in contrast to the results following exposure to the N₂ environment, where STM analysis demonstrated that the Cl–Si remained unchanged while the bare Si had observable defects attributed to water adsorption and oxidation. As Cl–Si has been previously demonstrated to be resistant to high levels of water vapor [11, 12], it is unlikely that ambient deterioration is simply a reaction with water. These and other studies also indicate that the oxidation of these and similar surfaces in ambient is complex, and likely involves a combination of adventitious species, such as C–O and O [13]. The observed ambient instability of Cl–Si aligns with previous literature studies. In one case, studies on UHV-prepared Cl–Si(111) exposure to ambient showed that it had a small amount of degradation after 10 min and significant degradation after 20 h of exposure [17]. Similar results were observed on Cl–Si(111) and Cl–Si(100) surfaces made by exposing wet-chemically-prepared H–Si(111) to Cl₂ gas at elevated temperature. The resulting Cl–Si(111) surface exhibited a small amount of SiO_x after a few minutes of ambient exposure, and significant amounts after 1 h [11, 16]. Cl–Si(100) surfaces were found to degrade faster than Cl–Si(111) [16]. These results also align with our observations on Cl–Si(100), which exhibits rapid deterioration with ambient exposures.

The ambient stability order of the halogens to oxidation, with I–Si > Br–Si > Cl–Si aligns perfectly with the trend in Pauling electronegativities (χ) of the halogens, with I (χ = 2.66) > Br (χ = 2.96) > Cl (χ = 3.16). The properties of halogen–Si have previously been shown to trend with χ values, with computational studies on the work function [37] of Si showing the increase in work function tracked with increase in χ with F–Si > Cl–Si > Br–Si > I–Si > H–Si. The differences in electronegativity also reflect the well-known trend of their reactivity in substitution reactions [38], with Cl being the most reactive of this set of halogens. This is likely playing a role in the different reactivity with ambient molecules seen in this work [39]. Interestingly, while I–Si has a higher χ than H–Si (χ = 2.2), the I–Si stability is similar to H–Si, with virtually no oxidation after 8 h. This suggest that χ is not the sole determinator of stability. It is likely that the steric bulk of the I provides additional protection for the I–Si against impinging gas molecules making the I–Si stability similar to the H–Si which does not have the same steric protection due to the small size of H. These results also align with observations on halogen-passivated Ge nanowires, where the stability of the halogen–Ge surfaces followed the order of Cl < Br < I, with larger and less electronegative halogens providing better protection against oxidation [40].

3.4. Co-passivation for pattern preservation

The individual stability of each Si passivation layer shown here (H, Cl, Br, and I) poses an interesting solution to APAM pattern preservation outside of UHV environments when taken in concert with past demonstrations of halogen adsorption into dangling bond structures formed on H–Si [14, 41, 42]. Figure 4 demonstrates this concept with a sequentially patterned and halogen-exposed H–Si sample that resulted in a complementary resist of multiple, distinct halogen-passivated regions on an otherwise fully H-passivated surface. First, the Cl–Si region was fabricated by patterning a 100 nm × 100 nm box (upper left) using STM lithography on an H–Si surface, then transferring it to the preparation chamber and exposing it to Cl₂. Etched alignment marks on the Si sample were used to reposition the STM tip and readily locate the patterned area. The process was repeated for the Br–Si (upper right) and then the I–Si (lower left) regions, with sequential patterning and exposure to Br₂ and I₂, respectively. In addition, a bare Si box was patterned (lower right) and intentionally left unpassivated to serve as an experimental control for comparing the extent of reactions with various environments. Vacancy features (circled) were present on the initial H–Si surface and serve to verify that the same sample was used for both exposures.

Zoom In Zoom Out Reset image size

Figure 4(a) shows the co-passivated sample after exposure to an inert N₂ environment for 2 h. In direct correlation with the results presented in figure 1, we find that each of the halogen–Si regions were remarkably robust against the N₂ exposure and showed negligible signs of degradation. The most significant change in surface morphology occurred within the bare Si box, which rapidly oxidized, similar to the bare Si sample shown in figure 1. Within the halogen–Si regions we found no increase in Si vacancies, indicating that the patterned boxes were fully passivated, suppressing etching and roughening pathways while annealing at 325 °C to remove adventitious physisorbed species [28]. The noticeable contrast between the Cl–Si, Br–Si, and I–Si regions serves as positive chemical identification in agreement with prior local density of states (LDOS) measurements reported in literature [43], which revealed a differing state density (I > Br > Cl) at the scan bias (+2.0 V) used to obtain the images shown in figure 4. That study used the LDOS difference to distinguish dilute amounts of mixed halogen adsorbates on Si(100) [43], and we find it also holds true for large, fully passivated regions shown here.

The co-patterned sample was then exposed to the laboratory ambient for 1 h, and the resulting surface is shown in figure 4(b). The sample displays clear signs of degradation. As seen in figure 4(b), the Cl–Si region has changed dramatically and has taken on a similar appearance to the bare Si region. This is in agreement with XPS results shown in figure 3, in which clear signs of oxidation are found on Cl–Si and the bare Si sample after exposure to ambient conditions. Interestingly, despite the lack of an SiO_x peak in the XPS data after a 1 h ambient exposure, figure 4 demonstrates that the Br–Si region displays signs of oxidation and deterioration, while the I–Si region has largely remained intact. This co-patterning technique was repeated on a smaller scale with nominally 10 nm boxes with similar results (see SI). As these results demonstrate, a fully halogen-passivated APAM pattern can withstand forays into inert environments without significant degradation of the pattern or resist. However, I passivation provides the greatest protection against ambient exposure.

In conclusion, the capability to form atomically-precise halogen-passivated regions on an otherwise H-passivated sample has been previously limited to UHV [14, 41, 42]. The results presented here study the stability of these samples in non-UHV environments to extend the use of atomic precision patterning to applications outside of UHV. The stability investigations of H–Si, Cl–Si, Br–Si, and I–Si demonstrated each to be stable in N₂ for extended periods of time. However, exposure of Cl–Si to the laboratory ambient resulted in a deterioration of the surface structure after just 15 min. In contrast, Br–Si and I–Si were highly resistant to degradation in the laboratory ambient over the course of the 8 h study and are likely to remain robust for much longer times.

The environments and time points explored in this work represent the real-world scenarios encountered upon transferring samples between processing and analysis tools. Our results highlight and confirm the previously known necessity to maintain an inert environment when working with Cl–Si outside of UHV conditions making glovebox and Schlenk setups required for every step of the process to prevent oxidation. The comparatively superior stability of Br and I passivation layers enables the use of halogen-passivated Si in real-world processes such as recent work where UHV-prepared Br–Si was transported off-site for solvothermal reactions with hydrazine [36]. Additionally, the ability to prepare ambient stable H and halogen resists opens up the opportunity to integrate APAM feature sizes with selective area chemistries, such as boron-based monolayer doping, that cannot be performed in UHV [44]. Incorporation of advanced and selective chemistries [44–47] targeting H and halogen passivation layers will be the next step to utilization of atomic precision patterning toward building advanced devices and architectures. Additionally, the successful demonstration of Br–Si and I–Si surface stability outside UHV enables the exciting possibility of integrating APAM patterns with commercial device processing.

We thank Karen Gaskell for XPS measurements performed at the University of Maryland Surface Analysis Center. This work was supported in part by the Laboratory Directed Research and Development program at Sandia National Laboratories, a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the US Department of Energy's National Nuclear Security Administration under Contract DE-NA-0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the US Department of Energy or the United States Government.

The data that support the findings of this study are available upon reasonable request from the authors.